Optimal Posted-Price Mechanism in Microtask Crowdsourcing

Abstract

Posted-price mechanisms are widely-adopted to decide the price of tasks in popular microtask crowdsourcing. In this paper, we propose a novel postedprice mechanism which not only outperforms existing mechanisms on performance but also avoids their need of a finite price range. The advantages are achieved by converting the pricing problem into a multi-armed bandit problem and designing an optimal algorithm to exploit the unique features of microtask crowdsourcing. We theoretically show the optimality of our algorithm and prove that the performance upper bound can be achieved without the need of a prior price range. We also conduct extensive experiments using real price data to verify the advantages and practicability of our mechanism.

Rating Mechanisms for Sustainability of Crowdsourcing Platforms

“…While some crowdsourcing systems consistently attract high-quality contributors, other seemingly similar ones suffer from low quality work, or even fail due to too little participation [9]. As contributors in crowdsourcing are mostly self-centered, it is of great importance to understand how to provide incentives for contributors to provide high-quality work, which has been studied for decades by a variety of work from different domains [10,12,18].…”

“…The types of reward (or compensation) used as incentive for contributors varies depending on the particular application. Some crowdsourcing systems (e.g., MTurk [6]) offer financial incentives for participation [18], many others are driven by social-psychological rewards, e.g., both intrinsic motivators like interest [10,12,20] or the satisfaction of benefiting a cause [9] (e.g., participating in a scientific research [13,40]), as well as extrinsic social rewards such as reputation or status [1]. There is now a growing effort in social psychology that aims to address what motivates contributors in crowdsourcing systems [2,21,23,24].…”

“…Crowdsourcing systems enable new types of problem solving and creative production across a range of domains, from citizen science [13] to product development [32], entertainment and the arts [28]. Individuals involved in crowdsourcing are typically heterogeneous, self-selected, and voluntary participants engaged in temporary, decentralized problem-solving activities, for a number of selfish and altruistic reasons [18]. In particular, these systems offer individuals opportunities for learning, social mobility and global economy, while enabling work that may be too difficult for machines, or too large a scale for small groups [7].…”

“…Accordingly, thoughtful design of incentivization schemes is critically important. A number of recent studies have demonstrated the power of incentive schemes [18,36,37] to elicit fair and accurate ratings and productivity in crowdsourcing, particularly when most aspects of performance are measurable. However, these approaches tend to favor task requesters over contributors, in that they are designed to enhance contributors' productivity and throughput.…”

“…However, these approaches tend to favor task requesters over contributors, in that they are designed to enhance contributors' productivity and throughput. For instance, in platforms offering micro-tasks, several incentive schemes (e.g., [18,37]) provide higher compensation to contributors with positive historical records. Although intuitive, these approaches may eventually erode the platform's sustainability over time, as they discourage contributors with less experience from participating (and therefore gaining experience).…”